The wide variation in the genomic base compositions of bacteria, which ranges among species from 14% to 75% G+C, has led to much speculation as to its origin, its persistence and its implications for bacterial fitness. Several adaptive explanations have been proposed~ however, the most persuasive and widely held view, first posited over 50 years ago, is that this variation is neutral, caused by differences among taxa in the underlying pattern of mutations. This view was recently countered through comprehensive sequence comparisons, which determined that mutations are universally biased towards A+T, even in bacteria with high genomic G+C contents, indicating a role of natural selection in determining base composition. Moreover, preliminary experiments revealed that bacteria expressing GC-rich versions of genes display higher growth rates than those expressing the identical protein from AT-rich versions, again showing that selection favors higher genic G+C in a genome where mutations are biased towards A+T. Why and how selection operates on the base composition of genes is unknown~ and none of the processes known to exert constraints on synonymous sites, or on gene and protein expression, can explain the selective force favoring high G+C. The goals of the proposed work are to determine (i) the source of this newly discovered selective process, (ii) the universality of its action, and (iii) the mechanism by which it operates. These goals will be accomplished in four Aims, the first of which employs synthetic gene constructs that are designed to identify the key compositional features of genes that foster the selective differences. In the second Aim, these gene constructs will be tested for their selective effects in bacterial groups that differ with respect to their mutational patterns and overall genomic base compositions~ results will reveal if selection acts on a cellular process that is common to all bacteria. The third Aim will examine effects of base compositional variation in native genes to determine the extent to which the selective force on genic G+C is physiologically relevant and has affected genes that have resided in a genome for millions of years. Although sequence comparisons and fitness assays both offer robust methods for detecting the past action of selection, they provide little information about its actual causes. Therefore, the final Aim will apply a sensitive, genome-wide approach for studying the interactions between gene transcripts and ribosomes to elucidate the mechanistic basis of compositional selection. Together, these studies will resolve a longstanding question in bacterial evolution and increase our understanding of the molecular evolutionary processes that shape bacterial genomes.